/polyscope-py

Python bindings for Polyscope

Primary LanguagePythonMIT LicenseMIT

polyscope-py

Python bindings for Polyscope. https://polyscope.run/py

Linux macOS Windows PyPI Conda

This library is a python wrapper and deployment system. The core library lives at https://github.com/nmwsharp/polyscope. See documentation at https://polyscope.run/py.

To contribute, check out the instructions here.

Installation

python -m pip install polyscope

or

conda install -c conda-forge polyscope

polyscope-py should work out-of-the-box on any combination of Python 3.5-3.12 and Linux/macOS/Windows. Your graphics hardware must support OpenGL >= 3.3 core profile.

For developers

This repo is configured with CI on github actions.

  • By default, all commits to the main branch build & run tests. Use [ci skip] to skip this.
  • Tagging a commit with [ci build] causes it to also build all precompiled wheels on a matrix of platforms to ensure the build scripts succeed.
  • Tagging a commit with [ci publish] causes it to build all precompiled wheels on a matrix of platforms AND upload them to pypi index

Deploy a new version

  • Commit the desired version to the master branch. Use the [ci build] string in the commit message to trigger builds, which should take about an hour.

  • Watch the github actions builds to ensure all wheels build successfully. The resulting binaries will be saved as artifacts if you want try test with them.

  • While you're waiting, update the docs, including the changelog.

  • Update the version string in setup.py to the new version number. When you commit, include the string [ci publish], which will kick of a publish job to build wheels again AND upload them to PyPI.

  • If something goes wrong with the build & publish, you can manually retry by pushing any new commit with "[ci publish]" in the message.

  • Create a github release. Tag the release commit with a tag like v1.2.3, matching the version in setup.py

  • Update the conda builds by committing to the feedstock repository. This generally just requires bumping the version number and updating the hash in meta.yml. Since meta.yml is configured to pull source from PyPi, you can't do this until after the source build has been uploaded from the github action.